Processing of Simulated Data

Fig. 4 shows a simulated original signal which consists of a random collection of clouds and impulses. The clouds are created by threshholding a superposition of two arrays of strongly correlated and weakly correlated random numbers so that a pre-specified fraction of the image remains black. The stars are simulated as a random collection of impulses which are distributed randomly over the image. The measured point spread function from the data file F372M.FITS, which was used to generate the simulated measured data, is also displayed.

Fig. 5 shows the simulated measured data together with the restored image. The noise is white and Poisson distributed with an SNR of 10 dB. The degree of blurring as well as the noticeable amount of noise is clearly visible. The restoration of the clouds is excellent, while the stars remain somewhat blurred. The reason for this difference is the use of the power density spectrum of the measured data which overestimates the correlation inherent in the data.